|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Sioen; |
|
4
|
|
|
|
|
5
|
|
|
use Sioen\HtmlToJson\Converter; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* Class HtmlToJson |
|
9
|
|
|
* |
|
10
|
|
|
* Converts html to a json object that can be understood by Sir Trevor |
|
11
|
|
|
* |
|
12
|
|
|
* @version 1.1.0 |
|
13
|
|
|
* @author Wouter Sioen <[email protected]> |
|
14
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php MIT |
|
15
|
|
|
*/ |
|
16
|
|
|
final class HtmlToJson |
|
17
|
|
|
{ |
|
18
|
|
|
/** @var array */ |
|
19
|
|
|
private $converters = array(); |
|
20
|
|
|
|
|
21
|
3 |
|
public function addConverter(Converter $converter) |
|
22
|
|
|
{ |
|
23
|
3 |
|
$this->converters[] = $converter; |
|
24
|
3 |
|
} |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Converts html to the json Sir Trevor requires |
|
28
|
|
|
* |
|
29
|
|
|
* @param string $html |
|
30
|
|
|
* @return string The json string |
|
31
|
|
|
*/ |
|
32
|
3 |
|
public function toJson($html) |
|
33
|
|
|
{ |
|
34
|
|
|
// Strip white space between tags to prevent creation of empty #text nodes |
|
35
|
3 |
|
$html = preg_replace('~>\s+<~', '><', $html); |
|
36
|
3 |
|
$document = new \DOMDocument(); |
|
37
|
|
|
|
|
38
|
3 |
|
libxml_use_internal_errors(true); |
|
39
|
|
|
// Load UTF-8 HTML hack (from http://bit.ly/pVDyCt) |
|
40
|
3 |
|
$document->loadHTML('<?xml encoding="UTF-8">' . $html); |
|
41
|
3 |
|
$document->encoding = 'UTF-8'; |
|
42
|
|
|
|
|
43
|
|
|
// fetch the body of the document. All html is stored in there |
|
44
|
3 |
|
$body = $document->getElementsByTagName("body")->item(0); |
|
45
|
|
|
|
|
46
|
3 |
|
$data = array(); |
|
47
|
|
|
|
|
48
|
|
|
// loop trough the child nodes and convert them |
|
49
|
3 |
|
if ($body) { |
|
50
|
3 |
|
foreach ($body->childNodes as $node) { |
|
51
|
3 |
|
$data[] = $this->convert($node); |
|
52
|
3 |
|
} |
|
53
|
3 |
|
} |
|
54
|
|
|
|
|
55
|
3 |
|
return json_encode(array('data' => $data), JSON_UNESCAPED_UNICODE); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Converts one node to json |
|
60
|
|
|
* |
|
61
|
|
|
* @param \DOMElement $node |
|
62
|
|
|
* @return array |
|
63
|
|
|
*/ |
|
64
|
3 |
|
private function convert(\DOMElement $node) |
|
65
|
|
|
{ |
|
66
|
3 |
|
foreach ($this->converters as $converter) { |
|
67
|
3 |
|
if ($converter->matches($node)) { |
|
68
|
3 |
|
return $converter->toJson($node); |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
} |
|
72
|
|
|
} |
|
73
|
|
|
|
WouterSioen /
sir-trevor-php
